theSkyNet POGS project Update
log in

Advanced search

Message boards : News : theSkyNet POGS project Update

1 · 2 · Next
Author Message
Profile alexbeckley
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Jan 13
Posts: 345
Credit: 5,696,421
RAC: 0
Message 3591 - Posted: 8 Dec 2014, 2:47:51 UTC
Last modified: 12 Dec 2014, 0:54:32 UTC

Hi All,

We’re planning an upgrade for POGS in the next few months that will see us add more calculations to the suite already conducted on each galaxy image, add more data types (such as higher resolution images from the WISE telescope, http://www.nasa.gov/mission_pages/WISE/main/, which collects infrared light) and much, much more.

There are some big changes, and some smaller ones, included in the upgrade. We’ve given a short summary of everything below, but if you’d like any more specific detail or have questions then please let us know.

As part of the upgrades we intend to run some simulated galaxies through the upgraded POGS code as a check on the accuracy of our the data processing. These 'fake' galaxy images are created by supercomputers as part of a large cosmological simulation http://www.illustris-project.org. Once the simulated galaxies are processed by the upgraded POGS ‘spectral energy distribution’ fit code, we can compare the results to the known properties used when creating the fake (but highly realistic) galaxies to confirm the upgraded code works as expected and is producing accurate science results for the astronomers.

We’re also upgrading the data retrieval pipeline, continuing to get enhanced infrared images from the WISE telescope for all galaxies that have already been processed, as well as some more galaxy data from other new sources. This new data will broaden our results on the galaxies that have been already processed - helping the astronomers to see more of the bigger picture, so to speak.

Kevin and the team will also upgrade the input data processing pipeline as well. This will include improving the pre-processing of input images to enhance their usefulness in especially faint regions (like the outer parts of a galaxy). A technique called adaptive smoothing will be used to achieve this result without blurring of bright regions. They will also develop ‘parameter prior cubes’, sets of position-dependent bracketing parameter values that should be used in SED fitting calculations. You can think of them as educated guesses for each parameter. This will stop us calculating every possible combination of parameters on every galaxy — saving time and resources. This also means we will be changing the credit system from the current points-per-pixel based model, as with ‘priors’ all pixels are no longer equal. We will let you know when the credit change will occur and will do our best to make sure it’s equitable and results in approximately the same amount of credit you currently earn.

We will also be upgrading the output processing to produce better results images. This will include things like stellar masking (from catalogs, or directly from the images) to remove foreground stars from the calculations, and we also start using a code called ‘galfit’ (http://users.obs.carnegiescience.edu/peng/work/galfit/galfit.html) to give a mass surface density of the galaxies (a new calculation) and also try another code called ‘CIGALE’ (http://cigale.lam.fr/) and others on the data.

All of this comes together to mean better science output from POGS - something we’re really keen on and we hope you are too!

--TheSkyNet Team

Rapture
Avatar
Send message
Joined: 3 Aug 12
Posts: 26
Credit: 697,593
RAC: 0
Message 3594 - Posted: 8 Dec 2014, 21:19:12 UTC - in response to Message 3591.

Thanks for the wonderful update! Do you have any news regarding possible new projects in the pipeline that will run alongside with POGS?

Bill

Profile Dingo
Avatar
Send message
Joined: 3 Aug 12
Posts: 121
Credit: 23,101,786
RAC: 5,861
Message 3595 - Posted: 8 Dec 2014, 21:30:12 UTC - in response to Message 3594.

Thanks for the update from myself and BOINC@AUSTRALIA. Look forward to crunching POGS in the future and contributing as much as we can to the science.
____________

Proud Founder and member of



Have a look at my WebCam

Kirsten
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Avatar
Send message
Joined: 12 Sep 13
Posts: 48
Credit: 4,350,677
RAC: 0
Message 3597 - Posted: 9 Dec 2014, 3:15:23 UTC - in response to Message 3594.

Hi Bill,
No specific news on NEW projects to run alongside POGS but I'll be sending out an update this week hopefully with news about theSkyNet SourceFinder's port to BOINC. We're hoping to have that alpha testing in Jan/Feb 2015 and in beta by the end of Feb.
____________
Member of theSkyNet team.

Rapture
Avatar
Send message
Joined: 3 Aug 12
Posts: 26
Credit: 697,593
RAC: 0
Message 3599 - Posted: 9 Dec 2014, 21:07:43 UTC - in response to Message 3597.

Thanks! I am looking forward to your update soon!

Bill

Ringold
Send message
Joined: 10 Nov 13
Posts: 3
Credit: 1,163,077
RAC: 0
Message 3604 - Posted: 11 Dec 2014, 0:17:16 UTC - in response to Message 3591.

I was hoping for a little more detail on the upgraded code; sounds like that's roughly what the community would call "optimized apps"? There's also been past mention of a GPU app, as well.

More detail on the upgraded code would be cool no matter what exactly it refers to, from the perspective of my inner geek.

Jozef J
Send message
Joined: 12 Oct 13
Posts: 18
Credit: 45,235,815
RAC: 53
Message 3607 - Posted: 12 Dec 2014, 16:34:43 UTC - in response to Message 3591.

There are some big changes....

Means to support GPU? Ati or nvidia?
:-)

Kirsten
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Avatar
Send message
Joined: 12 Sep 13
Posts: 48
Credit: 4,350,677
RAC: 0
Message 3612 - Posted: 17 Dec 2014, 3:20:45 UTC - in response to Message 3604.

Hi Ringold, what more detail are you specifically looking for? The code will be a newer version of the current code that's used to calculate results on the galaxy pixels (magphys) and also a move to a virtual machine so we can include other codes easily, like galfit and CIGALE that are mentioned in the update. The links int eh update for those codes are an interesting intro to what they do (but are pitched at an astronomy audience, so may not be easily digestable!)

Hope that helps, let me know if you want even more :)
____________
Member of theSkyNet team.

Kirsten
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Avatar
Send message
Joined: 12 Sep 13
Posts: 48
Credit: 4,350,677
RAC: 0
Message 3613 - Posted: 17 Dec 2014, 3:21:15 UTC - in response to Message 3607.

Hi Jozef J,
No GPU announcment yet. Sorry!
____________
Member of theSkyNet team.

Matt
Send message
Joined: 26 Sep 13
Posts: 22
Credit: 1,091,858
RAC: 20
Message 3626 - Posted: 23 Dec 2014, 18:31:08 UTC - in response to Message 3591.

Hope they run on my old HTC Desire HD phone, unlike recent projects for 9 months, It's got a solid processor you can still use but the jobs are too big now!

Matt.

Profile alexbeckley
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Jan 13
Posts: 345
Credit: 5,696,421
RAC: 0
Message 3629 - Posted: 23 Dec 2014, 23:50:10 UTC - in response to Message 3626.

Hey Matt,

Agreed processing on a phones presents an amazing untapped potential for processing power however they require a different way of thinking. Most boinc projects simply port their existing applications to android this results in work units that take unreasonably long on most android devices. One possibility might be to handle auxiliary tasks such as post processing of results using android devices.

Cheers
Alex
____________
Cheers
Alex
theSkyNet.org webMaster
ICRAR

Matt
Send message
Joined: 26 Sep 13
Posts: 22
Credit: 1,091,858
RAC: 20
Message 3635 - Posted: 24 Dec 2014, 16:50:08 UTC - in response to Message 3629.
Last modified: 24 Dec 2014, 16:52:53 UTC

What's wrong with thinking laterally. Do it! I've got an old phone a spare power point at work, make my old phone mean something!

Jozef J
Send message
Joined: 12 Oct 13
Posts: 18
Credit: 45,235,815
RAC: 53
Message 3661 - Posted: 3 Jan 2015, 21:33:52 UTC

Pogs is actually a project that we are taking pictures for as long forgotten and unreal universe. something that is no longer realistic to how we compute it. light of those images of the galaxy to reach us, how long before than it was real?
It's not unnecessarily??........................

Profile Crunch3r
Send message
Joined: 16 Aug 12
Posts: 1
Credit: 10,304,018
RAC: 0
Message 3662 - Posted: 3 Jan 2015, 22:06:10 UTC - in response to Message 3612.

...and also a move to a virtual machine so we can include other codes easily, like galfit and CIGALE that are mentioned in the update...


I truly hope that you reconsider that. I'm not willing to install vbox on any of my machines since most of them are running vmware, Hyper-V...

Besides that, i do not see a real reason to do that in the fist place.

Would you mind explaining what drove you to the decision to use VMs ?

I'd truly hate to abandon POGs, but i'm not willing to install VBOX at all. :(
____________

Jozef J
Send message
Joined: 12 Oct 13
Posts: 18
Credit: 45,235,815
RAC: 53
Message 3663 - Posted: 3 Jan 2015, 23:02:11 UTC

I'll probably talk behind others but this is not good work from the science unfortunately.
last contact connected with them is of 17 and 23 December
probably they celebrate on the beach in sydney now ..

Science is the first place!!

AbidNibBeE
Send message
Joined: 23 Dec 14
Posts: 2
Credit: 503,388
RAC: 0
Message 3664 - Posted: 4 Jan 2015, 2:57:33 UTC

Hello,


This is a touch off topic, but I recently found out why my GPU eats the Milyway@home credits for lunch. It allows double precision floating point calculations. The vast majority of retail GPU's do not. I go through them in 15-30 seconds consistently. Where the POGS take an hour plus. I understand GPU crunching is on the way, but please consider allowing double precision floating point calculations like MilkyWay. The AMD tahiti chips will do it natively, or so I am told.

Profile alexbeckley
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Jan 13
Posts: 345
Credit: 5,696,421
RAC: 0
Message 3670 - Posted: 6 Jan 2015, 0:14:22 UTC - in response to Message 3663.

I'll probably talk behind others but this is not good work from the science unfortunately.
last contact connected with them is of 17 and 23 December
probably they celebrate on the beach in sydney now ..

Science is the first place!!


Hey Jozef,

Apologies for the lack of replies over the summer brake, but as you guessed we have been away on our summer/Christmas holidays but I can safely say we are back now and communications will continue as normal.
____________
Cheers
Alex
theSkyNet.org webMaster
ICRAR

Profile alexbeckley
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Jan 13
Posts: 345
Credit: 5,696,421
RAC: 0
Message 3671 - Posted: 6 Jan 2015, 0:17:53 UTC - in response to Message 3662.

...and also a move to a virtual machine so we can include other codes easily, like galfit and CIGALE that are mentioned in the update...


I truly hope that you reconsider that. I'm not willing to install vbox on any of my machines since most of them are running vmware, Hyper-V...

Besides that, i do not see a real reason to do that in the fist place.

Would you mind explaining what drove you to the decision to use VMs ?

I'd truly hate to abandon POGs, but i'm not willing to install VBOX at all. :(


I won't be able to explain this as well as kevin. How ever the short answer for using VBOX is the ability to use a greater number of science libraries. A lot of radio astronomy computing tools are written only for *nix systems and often require complicated setup processes to get them running. Thus they need to be deployed in a virtual box to allow them to run on the vast number of different machines that connect to BOINC.

We are well aware of the problems this causes to boinc clients that allready run on VMware and we are looking at what projects we would be able to run without using a VBox alongside the other projects.
____________
Cheers
Alex
theSkyNet.org webMaster
ICRAR

Profile Ubdaddy
Send message
Joined: 1 Aug 12
Posts: 45
Credit: 5,303,102
RAC: 2
Message 3680 - Posted: 6 Jan 2015, 20:53:34 UTC - in response to Message 3671.

Hi Alex,

When the time comes for the VBOX please publish a detailed How-To instructions.
I tried several times until I gave up to start crunching for LHC but to no avail.

Regards,

Yair

Profile alexbeckley
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Jan 13
Posts: 345
Credit: 5,696,421
RAC: 0
Message 3683 - Posted: 9 Jan 2015, 0:08:24 UTC - in response to Message 3680.

Sounds good, it's on the cards to do up some more detailed how-to's as it is :)

1 · 2 · Next

Message boards : News : theSkyNet POGS project Update


Main page · Your account · Message boards


Copyright © 2017 The International Centre for Radio Astronomy Research